Differential Impact of a Dutch Alcohol Prevention ProgramTargeting Adolescents and Parents Separatelyand Simultaneously: Low Self-Control and LenientParenting at Baseline Predict Effectiveness

Ina M. Koning & Jacqueline E. E. Verdurmen &

Rutger C. M. E. Engels &

Regina J. J. M. van den Eijnden &

Wilma A. M. Vollebergh

Published online: 2 February 2012# The Author(s) 2012. This article is published with open access at Springerlink.com

Abstract To test whether baseline levels of the factorsaccountable for the impact of the Prevention of Alcoholuse in Students (PAS) intervention (self-control, perceivedrules about alcohol and parental attitudes about alcohol),moderate the effect of the intervention. A cluster random-ized trial including 3,490 Dutch early adolescents (M age012.66, SD00.49) and their parents randomized over fourconditions: 1) parent intervention, 2) student intervention,3) combined intervention and 4) control group. Moderatorsat baseline were used to examine the differential effects ofthe interventions on onset of (heavy) weekly drinking at 34-month follow-up. The combined intervention was only effec-tive in preventing weekly drinking among those adolescentswho reported to have lower self-control and more lenientparents at baseline. No differential effect was found for theonset of heavy weekly drinking. No moderating roles of self-control and lenient parenting were found for the separatestudent and parent interventions regarding the onset of

drinking. The combined intervention is more effectiveamong adolescents with low-self control and lenient parentsat baseline, both factors that were a specific target of theintervention. The relevance of targeting self-control in adoles-cents and restrictive parenting is underlined.

Keywords Early adolescents . Alcohol use . Intervention .

Randomized trial . Moderation

Introduction

In recent years, the number of Dutch adolescents drinking ata young age has increased (Monshouwer et al. 2008). At thesame time more knowledge of the risks involved in under-age drinking has become available (Behrendt et al. 2009;Brown et al. 2008). Although the literature shows mixedfindings on the effectiveness of preventive alcohol interven-tions (Foxcroft et al. 2002; Spoth et al. 2008b), some inter-ventions have proved to significantly affect the drinkingbehavior of adolescents. However, interventions may havedifferent effects on different subgroups of adolescents.Thus, a relevant question is whether these effects apply toadolescents in general.

So far, most studies examining moderation effects focusedon demographic moderators, such as gender, level of educa-tion and ethnicity, or on the initial level of outcome of interest(e.g., Brown et al. 2005; Sloboda et al. 2009; Spoth et al.2006). For example, although most studies find more favor-able effects of alcohol interventions in adolescents whodrink more frequently at baseline (Brown et al. 2005; Spothet al. 2008a), not much support for beneficial effects in other

I. M. Koning (*) : R. J. J. M. van den Eijnden :W. A. M. VolleberghDepartment of Interdisciplinary Social Science,Utrecht University,P.O. Box 80.140, 3508 TC Utrecht, the Netherlandse-mail: i.koning@uu.nl

I. M. Koning : J. E. E. VerdurmenTrimbos Institute,Netherlands Institute of Mental Health and Addiction,P.O. Box 725, 3500 AS Utrecht, the Netherlands

R. C. M. E. EngelsBehavioural Science Institute, Radboud University Nijmegen,P.O. Box 9104, 6500 HE Nijmegen, the Netherlands

Prev Sci (2012) 13:278–287DOI 10.1007/s11121-011-0267-9

higher-risk groups (e.g., boys and adolescents in lowereducation) has been found (e.g., Spoth et al. 2006; Trudeauet al. 2003). However, it is particularly relevant to analyzewhether the theoretically relevant factors that are targeted inthe intervention also act as potential moderator variables.

Examination of the moderation effects of intervention-induced factors is important for two reasons. First, it pro-vides theoretical evidence for the justification of targetingthese specific factors in interventions. Second, insight intosubgroups who have more favorable effects from an inter-vention helps in finding the best target groups for imple-menting this intervention (Kraemer et al. 2002), especiallywhen these subgroups can be identified by demographicvariables. Thus, it is relevant to examine the moderatingrole of the intervention-induced factors on the effect of theintervention condition on outcome of interest.

Recently, a Dutch school-based multi-component preven-tion program (PAS; Prevention of Alcohol use in Students)was shown to effectively postpone the onset of drinking by10% when early adolescents as well as their parents weretargeted (Koning et al. 2009), whereas targeting adolescentsor parents separately revealed no significant effects. Theadolescent part of the intervention was developed with aview to increasing self-control and fostering healthy atti-tudes in adolescents. This target behavior was chosen be-cause the inability to refuse alcohol (low self-control) inadolescents appears to be a powerful predictor of (onsetof) alcohol use in adolescents (e.g., Wills et al. 2001).Therefore, increasing the level of self-control in adolescentsis often targeted in alcohol interventions, in many cases suc-cessfully (Foxcroft et al. 2002; Tobler et al. 2000). In theparent intervention, parents were encouraged to maintainrestrictive rule setting and attitudes about alcohol use. Thistarget behavior was chosen on the basis of a number ofstudies that consistently revealed strong effects of restrictiveparenting (rules and attitudes about alcohol) on adolescentdrinking (Koning et al. 2010b; Spijkerman et al. 2008; Vander Vorst et al. 2006, 2007; Yu 2003). That is, adolescentswith stricter parents (e.g., who prohibit drinking) are lesslikely to drink alcohol (heavily). In addition, interventionstargeting parenting behavior in order to induce change intheir offspring’s drinking have shown very promising results(Smit et al. 2008; Spoth et al. 2008b).

The combined intervention considered in this studyrevealed that this intervention indeed modified the theory-based factors as hypothesized. By increasing self-controlamong adolescents and by enhancing strict rule enforcementand negative attitudes in parents about adolescent alcoholuse, the onset of weekly drinking was postponed (Koning etal. 2010a). Koning et al. (2009) showed that only when bothadolescents and their parents were targeted was the onset ofdrinking postponed, whereas the separate parent or adoles-cent interventions were not effective. This makes us wonder

if the combined intervention is differentially effective fordifferent levels of self-control and restrictive parenting. Inaddition, the separate interventions may become effective ifhigh levels of self-control in adolescents and perceived strictparenting are present at baseline. That is, the adolescentintervention may become effective when parents set restric-tive rules and have strict attitudes, while the parent inter-vention may become effective among those adolescentswith a high level of self-control.

According to the risk moderation hypothesis, an inter-vention should be more effective among high-risk groupsthan among moderate- and low-risk groups. Although thisdoes not seem to account for moderation of demographicfactors, it may apply to intervention-induced factors. Ado-lescents who lack the intervention-targeted protectivebehaviors can be considered among the high-risk groups.With respect to the PAS intervention, this involves adoles-cents with low self-control and low restrictive parents atbaseline. Therefore, it can be expected that specificallythose higher-risk groups of adolescents benefit more fromthe intervention.

As the student and parent interventions specifically targetadolescent self-control and restrictive parenting respective-ly, more change can be induced in groups that are low at thestart of the intervention in comparison to adolescents withhigh levels of these behaviors at baseline. Increasing self-control in adolescents is often a target of alcohol interven-tions, in many cases successfully (Foxcroft et al. 2002;Tobler et al. 2000). Moreover, several studies point to theimportance of testing self-control among adolescents as amoderator in intervention trials (e.g., Brown et al. 1998;Demmel et al. 2004). Earlier studies have confirmed that alower level of self-control at baseline indeed appeared to berelated to a higher degree of change in the desired directionduring the intervention (Brown et al. 1998; Demmel et al.2004). In addition, Brody et al. (2008) showed that the effectof their parenting program on conduct problems in adoles-cents was greater for youth with lower levels of self-control.Yet, to our knowledge, no data are available with respect tothe moderation of restrictive parenting. Thus, a relevantquestion is to investigate whether the combined interventionis indeed most effective in the adolescent groups that arelow in self-control and perceived rule setting at baseline andin parents with a tolerant attitude with respect to adolescentdrinking.

Current Study

In this study, we examine whether the degree to whichbaseline levels of factors accountable for the effect of thecombined PAS intervention (i.e., self-control, perceived pa-rental rules and parental attitudes about alcohol use) mod-erate the effects of the different intervention conditions

Prev Sci (2012) 13:278–287 279

(parent and student intervention separately and simulta-neously). In addition, in order to detect the subgroups onwhom the intervention has more favorable effects, the char-acteristics of these subgroups have to be identified. Demo-graphic factors that have been found to relate to low self-control in adolescents or permissive parenting are gender(Sumter et al. 2009), age (Sumter et al. 2009; Van der Vorstet al. 2005), level of education (Spijkerman et al. 2008) andreligion (Spijkerman et al. 2008). In a cluster randomizedtrial, the effects were measured for onset of both weekly andheavy weekly drinking at 34-month follow-up in a sampleof 3,417 adolescents and their parents.

Method

Procedure and Participants

In April 2006, 80 schools were randomly selected from thelist of all public secondary schools in the Netherlands, andwere requested to participate in the study. A total of 20schools were willing to participate.

Both the students and their parents were involved in thisstudy. Student data were collected in classrooms by onlinequestionnaires, available from a secure website. Researchassistants were trained to administer the survey. Ques-tionnaires for parents were sent to their home address,along with a letter of consent. Non-responding parentswere encouraged to participate by a written reminderand phone call. The parent who completed the firstquestionnaire was asked to complete the subsequentquestionnaires as well. In most cases (83%), the respondingparent was female. Both parental and student data weregathered in September/October 2006, before any interven-tion was carried out, and again 34 months later in June/July2009. For a more detailed description of the procedure, seeKoning et al. (2009).

Randomization

The participating schools were randomly assigned by anindependent statistician to one of the following conditions:1) parent intervention, 2) student intervention, 3) both par-ent and student intervention (combined intervention), or 4)control condition consisting of the regular curriculum. Ran-domization was carried out centrally, using a blocked ran-domization scheme (block size 5) stratified by level ofeducation, with the schools as the unit of randomization.Within each participating school, all first-year students par-ticipated in the intervention. After randomization, oneschool refused further participation for reasons unrelated tothe study. This school was originally randomized to thecontrol condition.

Interventions

Parent Intervention (PI) This intervention targets parentalrules for their children’s alcohol use. The intervention wasmodeled on a Swedish intervention, The Örebro Preventionprogram (for details, see Koutakis et al. 2008). The inter-vention was carried out at the first parents’ meeting at thebeginning of each school year (September/October 2006and 2007), in which other school-related topics were alsodiscussed. A brief presentation was given at the plenarymeeting by a social worker who was trained by the NationalInstitute of Mental Health and Addiction. Thereafter, parentsof children from the same class got together to reach con-sensus on a shared set of rules about alcohol use. Threeweeks later an information leaflet with a summary of thepresentation and a report of the outcome of the class meetingwas sent to parents’ home addresses.

Student Intervention (SI) The SI is the renewed digital alco-hol module of the Dutch prevention program ‘The HealthySchool and Drugs’ (HSD). The alcohol module makes useof e-learning and is based on principles of the theory ofplanned behavior (Ajzen and Fishbein 1990) and socialcognitive theory (Bandura 1996). In the interactive lessonsstudents were provided with information about alcohol andthe risks involved in drinking. Students were trained toincrease their self-control and develop healthy attitudestowards alcohol use by means of exercises such as shortmovies, texts following questions, animations and a chatroom to exchange opinions. Each lesson ends with a sum-mary task, such as creating a flyer or poster. The teachersreceived training in advance on how they could guide stu-dents in completing the lessons. In addition, informationwas given about how teachers could track the students whilecarrying out the intervention. The teachers conducted theintervention (four lessons) in all first-year classes in March/April 2007. A hard-copy booster session was provided bythe teachers 1 year later in March/April 2008. This boosterlesson on paper involved a repetition of the digital alcoholprogram. For example, propositions were put forward toencourage discussion and recall knowledge obtained fromthe digital program.

Control Condition (CC) Schools in the control conditionwere contracted not to start any alcohol-related interven-tions throughout the study period. However, as basicinformation about alcohol use is part of the standardcurriculum in the Netherlands, which involves the previ-ously supported message of learning to drink alcoholsafely instead of prohibiting it, they were allowed tocontinue this practice. The interventions were implementedand supervised by the National Institute of Mental Health andAddiction.

280 Prev Sci (2012) 13:278–287

Demographic Variables

Age (in years), gender, level of education and religion weredemographic variables taken into account as covariates and/or predictors. Age was measured by subtracting date ofresponse from the date of birth of the respondent. The levelof education reflects the type of education the adolescentengaged in. In the Netherlands, from the first year of sec-ondary school, the educational system consists of differenttypes of secondary education ranging from pre-vocational topre-university education. This variable was dichotomizedinto lower (pre-vocational and lower secondary education)versus higher (higher general secondary and pre-universityeducation) level of education. Religion was measured byasking if the adolescent was brought up religiously (‘Catholic,’‘Protestant,’ ‘Islamic,’ ‘Other, namely’ and ‘No religion’). Inthe current analyses, adolescents indicating ‘other religion’were recoded as missing (3.6%). Dummy variables werecomputed with ‘no religion’ as reference category.

Outcome Measures

The outcomes of interest for the proposed study were equalto the outcome measures used in the previous study in whichthe effectiveness of the intervention was described (Koninget al. 2009). In accordance with the registration of this studyin the Trial register (NTR649), we were interested in theeffect of the intervention on the incidence rate of onset ofdrinking. Therefore, dichotomous variables were computed.

The onset of weekly alcohol use was defined by aQuantity-Frequency measure (Engels and Knibbe 2000;Engels et al. 1999). To detect the incidence rate of onset of

weekly drinking, the scale was recoded into 0 0 ‘no weeklyuse’ and 1 0 ‘weekly user,’ if at least one glass of alcoholwas consumed on a weekly basis. This was only calculatedfor those adolescents who indicated a monthly prevalence ofdrinking. Onset of heavy weekly drinking was measured byasking how many glasses of alcohol the student usuallydrank on a weekend day (Engels et al. 1999). Boys drinkingfive or more glasses and girls drinking four glasses or moreper week were considered to be heavy drinkers. The scalewas recoded into a dichotomous variable with 0 0 ‘no heavyweekly drinking’ and 1 0 ‘heavy weekly drinking.’

Moderators

Self-control and restrictive parenting factors (rules and atti-tudes about alcohol) were moderators investigated in thisstudy. All moderators, except parental attitudes, werereported by the adolescent. Due to skewness of the data(see Table 1), all moderators were transformed intodummies (0/1 variables) on the basis of median split andwere used to compute interaction variables (cf. Spoth et al.2006).

Self-control measures the ability to control responses,to interrupt undesired behavioral tendencies and refrainfrom acting on them. The measure is the shorter versionof the original measure developed by Tangney et al.(2004). It consists of 13 items (alpha 0 .74) that were ratedon a 5-point scale, ranging from 1 ‘not at all like me’ to 5‘very much like me.’ For example, “I have trouble sayingno” and “I do certain things that are bad for me, if they arefun.” Items were reversely scored, higher scores indicatemore self-control. The sample was divided into 0 0 ‘high

Table 1 Descriptive statistics ofdemographic and moderatingvariables and alcohol use

SD standard deviation

%

Demographic variables

Religion

No religion 59.5Catholic 22.5

Protestant 13.6

Islamic 4.4

Gender

Boy 51.0

Level of education

Low 40.1

Alcohol use

Weekly drinkers 53.8Heavy weekly drinkers 21.2

Moderating variables Mean SD Skewness Kurtosis

Self-control (1–5) 3.65 0.50 −0.28 0.15

Rules about alcohol (1–5) 4.64 0.46 −2.04 6.06

Attitude toward alcohol (1–5) 4.59 0.50 −2.48 10.77

Prev Sci (2012) 13:278–287 281

self-control’ and 1 0 ‘low self-control’ based on the medianscore.

Rules about alcohol use reflect the degree of rule-settingbehavior by the parents, as experienced by the adolescent(Van der Vorst et al. 2005). Items included “I am allowed tohave one glass of alcohol when my parents are at home,” “Iam allowed to drink several glasses of alcohol when myparents are not home” and “I am allowed to drink alcoholwhen I am at a party with my friends.” It consisted of themean of ten items (alpha 0 .90) rated on a 5-point scale from1 ‘never’ to 5 ‘always’ reversely scored; i.e., higher scoresindicate more rule-setting behavior. The median split wasused to distinguish strict parents from lenient parents(scored as ‘0’ and ‘1’ respectively), as perceived by theadolescent.

Attitudes about alcohol use measure the degree to whichthe parent finds it acceptable (1 0 not at all acceptable to 5 0very acceptable) for a 12/13-year old adolescent to drinkalcohol in various situations (Brody et al. 1999; Van derVorst et al. 2006). Originally seven items were listed; in thisstudy we added one item (drink alcohol on a Saturdayevening with parents; alpha 0 .79). Responses were dichot-omized, so that parents with strict and lenient attitudes(scored as ‘0’ and ‘1’ respectively) could be distinguished.

Statistical Analyses

Data were analyzed in accordance with the intent-to-treatprinciple. All analyses were conducted with Stata/SE ver-sion 9.2. At follow-up, 13.8% of the data were missing,mainly due to students changing schools. Missing data werehandled by regressing imputation as implemented in Statawhile using rules and attitudes about alcohol, alcohol use byboth the students and their parents at baseline and age,gender, and level of education (lower vs. higher secondaryeducation) as predictors of outcome.

The randomization had resulted in a slightly unevendistribution across the active conditions compared with thecontrol condition in terms of age, gender, and level ofeducation. Therefore, the moderation analyses were con-ducted with these variables as covariates to control for anypossible bias stemming from the imbalance.

The cluster effect (introduced in the data since studentswere ‘nested’ in classes) was handled data-analytically bygetting robust variance-related estimates based on the first-order Taylor-series linearization method, using Stata’s pro-cedures for design-based analyses. We corrected for thecluster effects at class level, as the interventions were carriedout in classes. To test moderation effects, interaction varia-bles were computed between the moderator and the inter-vention dummies (received vs. did not receive theintervention). Furthermore, multiple logistic regression wasused to analyze the main and moderation effects. Post

estimation linear combination calculations, as implementedin Stata, were used to be able to interpret the significantmoderation effects.

Participants

A total of 3,490 students were asked to participate in thestudy. Of these, 122 students did not participate due to therefusal of their parents, or because they were absent fromschool the day the questionnaire was administered. Thisresulted in a response rate of 97% (N03,368) at baseline.It was required to include only students who did not meetthe criteria for weekly drinking at baseline, and were there-fore ‘at risk’ of becoming manifest as new cases of (heavy)drinking at follow-up. Therefore we excluded 431 students(12.7%) who already were weekly drinkers at baseline (n0306), or who responded inconsistently on the quantity andfrequency items measuring weekly drinking (n0125). Thisresulted in a total of 2,937 students and 81% of their parentseligible for analysis. Of this group, a total of 2,533 students(86.2%) and 63.3% of their parents stayed in the programand completed the follow-up assessment after 34 months.Intention-to-treat analyses were based on 2,937 students notmanifesting weekly drinking at baseline and 2,381 parents.

Characteristics of the Sample at Baseline

The total student sample had a mean age of 12.66 (SD0

0.49), consisting of 51% boys, and 40% in lower secondaryeducation (Table 1). Most students reported not having anyreligion (59.5%). An extensive description of the sample atbaseline for each condition can be found in Koning et al.(2009). Of all participating parents, most were female (81%)and between the age of 35 and 49 (90%). More than onethird of the parents had secondary education (38%), 34%vocational education and 27% third-level education.

Results

Onset of Weekly Drinking

Tables 2, 3 and 4 depict the main and interaction effects ofintervention conditions with self-control and rules and atti-tudes about alcohol use respectively on the onset of weeklydrinking.

Self-control in adolescents moderated the effect ofthe combined intervention on the onset of weekly drink-ing (ORinteraction00.62; 95% CI 0.40–0.96, p00.04). Postestimation calculations showed that the effect of the combinedintervention on weekly drinking was significant among ado-lescents with low self-control (OR00.52; 95% CI 0.36–0.75,p00.00), and not significant among adolescents with high

282 Prev Sci (2012) 13:278–287

self-control (OR00.84; 95% CI 0.56–1.27, p00.42). Thus,the combined intervention is only effective in adolescentswith low self-control and not in adolescents with high self-control at baseline (Fig. 1). No moderation effects of thestudent and parent interventions separately were found.

Rules about alcohol moderated the effect of the combinedintervention on the onset of weekly drinking (ORinteraction0

0.63; 95% CI 0.41–0.98, p00.04). Post estimation calcula-tions showed that the effect of the combined intervention onweekly drinking was significant among adolescents report-ing that their parents had lenient rules about alcohol use

(OR00.63; 95% CI 0.41–0.98, p00.00), and not significantamong adolescents reporting to have strict parents (OR00.84; 95% CI 0.59–1.19, p00.32). Thus, the combinedintervention is effective in adolescents with lenient parentsand not in adolescents with strict parents regarding alcoholuse (Fig. 2). No moderation effects of the student and parentinterventions separately were found.

Parental attitudes did not moderate the effect of any of theintervention conditions on the onset of weekly drinking.

Onset of Heavy Weekly Drinking

In addition to the direct effect of the combined intervention onthe onset of heavy weekly drinking, no moderation effects ofany of the intervention conditions were found. Thus, theeffects of the interventions on the onset of heavy weeklydrinking did not differ according to the level of self-control,perceived rule-setting and parental attitudes at baseline.

Additional Analysis

In order to be able to detect those subgroups of adolescentsthat respond most favorably to the combined intervention (alow level of self-control and low restrictive rules aboutalcohol), we performed logistic regression analyses withlevel of education, age, gender and religion (religion vs.no religion) at baseline as predictors, and level of self-control (low vs. high) and rules about alcohol (low vs. high)as measures of outcome while controlling for the clustereffect.

Being Islamic as opposed to having no religion reducedthe chance of having low self-control (OR00.47; 95% CI

Table 2 Interaction effects between intervention conditions and self-control on the onset of weekly drinking while controlling for age,gender and level of education

OR 95% CI

Main effects

Age 1.33 1.13–1.55

Gender (1 0 boy) 0.93 0.78–1.11

Level of education (1 0 low) 1.17 0.95–1.45

Parent intervention 1.00 0.65–1.54

Student intervention 1.02 0.69–1.51

Combined intervention 0.84 0.56–1.27

Self-control (1 0 low) 1.77 1.30–2.43

Interaction effects

Parent intervention x self-control 0.70 0.42–1.17

Student intervention x self-control 0.69 0.46–1.05

Combined intervention x self-control 0.62 0.40–0.96

OR Odds Ratio, CI Confidence Interval

ORs in italic are statistically significant (p<.05)

Table 3 Interaction effects between intervention conditions and rulesabout alcohol on the onset of weekly drinking while controlling forage, gender and level of education

OR 95% CI

Main effects

Age 1.37 1.17–1.61

Gender (1 0 boy) 0.93 0.78–1.11

Level of education (1 0 low) 1.20 0.97–1.48

Parent intervention 1.00 0.65–1.54

Student intervention 1.01 0.72–1.44

Combined intervention 1.01 0.71–1.44

Rules about alcohol (1 0 lenient) 2.50 1.79–3.48

Interaction effects

Parent intervention x rules 0.67 0.44–1.01

Student intervention x rules 0.72 0.46–1.13

Combined intervention x rules 0.63 0.41–0.98

OR Odds Ratio, CI Confidence Interval

ORs in italic are statistically significant (p<.05)

Table 4 Interaction effects between intervention conditions and atti-tude about alcohol on the onset of weekly drinking while controllingfor age, gender and level of education

OR 95% CI

Main effects

Age 1.38 1.16–1.64

Gender (1 0 boy) 0.95 0.79–1.15

Level of education (1 0 low) 1.18 0.94–1.47

Parent intervention 0.76 0.57–1.41

Student intervention 0.74 0.59–1.34

Combined intervention 0.57 0.52–1.19

Attitude toward alcohol (1 0 lenient) 1.20 0.56–1.23

Interaction effects

Parent intervention x attitude 1.18 0.68–2.06

Student intervention x attitude 1.19 0.71–2.02

Combined intervention x attitude 1.37 0.83–2.26

OR Odds Ratio, CI Confidence Interval

ORs in italic are statistically significant (p<.05)

Prev Sci (2012) 13:278–287 283

0.31–0.69, p0 .000), while being at a lower level of educa-tion (OR01.99; 95% CI 1.64–2.43, p0 .000) significantlyincreased the likelihood of having low self-control(Table 5).

Being Islamic had the only significant correlation withthe level of rule setting in parents. Adolescents of Islamicreligion (as opposed to no religion) have a lower risk ofreporting low levels of restrictive parenting than adolescentsreporting to have no religion (OR00.17; 95% CI 0.09–0.30,p0 .000).

Logistic regression analysis performed on adolescentsreporting both low self-control and low restrictive rulesetting (25% of the total sample) showed that, again,level of education and religion were significant correlatesof having low self-control and low restrictive parents.That is, adolescents in low education (OR01.61; 95%CI 1.31–1.97, p0 .000) are more likely and those of Islamicreligion (OR00.16; 95% CI 0.08–0.35, p0 .000) are lesslikely to have low self-control and to live with low restric-tive parents.

Discussion

In the present study we examined the moderating effects offactors causing the effectiveness of an alcohol interventiontargeting parents and adolescents simultaneously (PAS). Ofall moderators investigated in this study (i.e., self-control,perceived rule-setting and parental attitudes about alcohol),both the initial level of self-control among adolescents andperceived rule-setting by parents moderated the effects ofthe combined intervention on the onset of weekly drinking.The present findings revealed that the combined interven-tion is most effective in delaying the onset of weekly drink-ing among those adolescents who reported to have low self-control and lenient parents at baseline. This underscores therelevance of targeting self-control in adolescents and restric-tive parenting in their parents. No differential effects of theseparate parent and student intervention on delaying theonset of (heavy) weekly drinking were found.

The finding that adolescents who perceive their parents tobe more lenient towards alcohol use and who experience

% W

eekl

y D

rink

ers

Fig. 1 Percentages (unadjusted)of weekly drinkers with low andhigh self-control in thecombined and control condition

% W

eekl

y D

rink

ers

Fig. 2 Percentages (unadjusted)of weekly drinkers with lenientand strict parents in the combinedand control condition

284 Prev Sci (2012) 13:278–287

little behavioral control over their alcohol use benefit morefrom the combined intervention is in line with the riskmoderation theory. According to the risk moderation hy-pothesis, high-risk groups should benefit more from preven-tive alcohol interventions than moderate- and low-riskgroups. We defined high-risk groups in terms of the theo-retically relevant target behaviors of the intervention; i.e., asthose scoring low on the target behaviors at baseline. AsPAS targeted self-control and restrictive parenting, adoles-cents with a low level of self-control and low restrictiveparents were considered to be a high-risk group. Asexpected, lower levels of the behaviors targeted by PAS,resulted in more favorable effects of the intervention. Theimportance of the moderating role of self-control has alreadybeen demonstrated by Brown et al. (1998) in a sample ofadults participating in an abstinence-focused inpatient alco-hol treatment program. As far as we know, no studies areavailable that tested the moderating role of rules parents setabout alcohol, that were also a target of the alcohol inter-vention. This study is one of the first showing the relevanceof testing moderation of intervention-induced factors ofalcohol interventions targeting early adolescents and theirparents. In addition to the effects of PAS in the generalpopulation (Koning et al. 2009), the fact that the effectswere more favorable among the groups scoring low on theintervention-induced behaviors at baseline underlines theappropriateness of choosing self-control and restrictive par-enting as targets for the intervention.

We can only speculate about the lack of significance ofmoderation in respect of parental attitudes. Attitudes ofparents about alcohol use, as opposed to specific rules aboutalcohol, represent not so much concrete behavior as a state ofmind. It is possible that the moderating effect of the combinedintervention on the onset of (heavy) weekly drinking was notfound due to the, sometimes subconscious, cognitive accep-tance of parents not to agree with early drinking. On the other

hand, changing their behavior by setting rules requires moreeffort from parents and results in those parents who werelenient showing the highest rates of change.

The separate parent and student intervention parts did notbecome effective in subgroups of adolescents, based on theirinitial level of self-control and restrictive parenting. This mayindicate that the effects of the separate interventions were notstrong enough to foster change in onset of drinking even whenthe pertinent behaviors were low at baseline. However, theresults do suggest that the separate interventions tend to bemore beneficial for adolescents with low self-control andthose experiencing lenient parenting. But in order to signifi-cantly impact the onset of drinking in early adolescents,behaviors in both parents and adolescents should be targeted.These findings underscore the additional effect of the com-bined intervention over and above the separate interventions.

While a differential effect of the combined interventionwas found for the onset of weekly drinking, this was not thecase for the onset of heavy weekly drinking. That is, theeffect of the combined intervention on the onset of heavyweekly drinking did not differ across the level of self-control and strict parenting. We can only speculate as towhy the results are different for weekly and heavy weeklydrinking. It is possible that self-control and parenting aremostly related to the onset of less severe drinking behavior,due to the high variance within adolescents and parents. Inthis way the intervention can induce high rates of change ina large number of adolescents scoring low on these behav-iors. However, once adolescents drink heavily on a weeklybasis, the lower variance of self-control and strictness ofparents may explain the lack of differential effect in thesesubgroups. More research on the role of self-control andparenting in diverse drinking intensities could provide moreinsight into this issue.

Additional analyses showed that those at risk for having lowself-control and low restrictive parents are particularly found

Table 5 Logistic regression of self-control and rules about alcohol predicted by demographic factors at baseline (gender, level of education andreligion)

Self-control (1 0 low) Rules about alcohol (1 0 low) Self-control and rules about alcohol (1 0 low on both)

OR p-value OR p-value OR p-value

Gender (1 0 boy) 1.16 .06 0.96 .64 1.12 .18

Age 1.05 .54 0.89 .18 0.93 .45

Level of education (1 0 low) 1.99 .00 1.15 .14 1.61 .00

Religion (reference 0 no religion)

Catholic 0.89 .32 1.00 .99 0.91 .41

Protestant .98 .91 0.83 .15 0.94 .69

Islamic .47 .00 0.17 .00 0.16 .00

OR Odds Ratio

ORs in italic are statistically significant (p<.05)

Prev Sci (2012) 13:278–287 285

among the lower educated and non-Islamic adolescents. Thisfinding is in support of a previous study that demonstrated thatadolescents from families with a lower socioeconomic statuswere more likely to have more lenient parents (Spijkerman etal. 2008). In addition, as drinking alcohol is not accepted inIslamic cultures and therefore alcohol use is not very com-mon among immigrant groups (Monshouwer et al. 2003),parents of immigrant youth are expected to bemore strict aboutalcohol. On the other hand, adolescents in lower education areat higher risk for alcohol use (Salonna et al. 2008). Thus, thelevel of education and religion of the adolescent are signif-icant indicators for identifying the subgroups of adolescentswho benefit most from the combined intervention.

Study Limitations

The results of this study should be considered in the light ofsome limitations. First, the results are based on self-reportedquestionnaires. Although susceptible to social desirabilitybiases, self-reports on alcohol use are found to be a validmethod to assess alcohol use in adolescents (Del Boca andDarkes 2003; Koning et al. 2010c). Second, the findings arebased on one follow-up measurement only. Examining dif-ferential effects of the PAS intervention on alcohol use atlater follow-ups may demonstrate the strength of the currentresults. Third, one should be careful in generalizing thefindings to other countries with other drinking cultures.The Netherlands is considered to have a fairly lenient drink-ing culture, in which adolescents drink more frequentlyrelative to other European countries (Hibell et al. 2009).Therefore, replication of this study in other countries isnecessary in order to generalize these findings to countrieswith a stricter drinking culture. Fourth, although we man-aged to include the number of schools and participants thatwere needed to power the trial (see Koning et al. 2009), only25% of all schools that were approached actually participat-ed. No data are available on whether the participatingschools differ from the non-participating schools. Thismay affect the generalizability of our findings. Last, in thisstudy we included education, age, gender and religion aspredictors to detect the specific subgroups that the interven-tion may benefit the most. It should be noted, however, thatapart from these characteristics, other predictors could havebeen useful to include in the analyses, such as family situ-ation and problematic behavior.

Implications

The present and previous results of the PAS intervention(Koning et al. 2009) indicate the need for targeting adoles-cents as well as parents, over and above targeting either of

them separately in a school setting. It is suggested that theconcurrent increase of self-control in adolescents and restric-tive parenting in parents are behaviors that can be targetedeffectively by alcohol interventions. Prevention workers andschools wanting to delay the onset of weekly drinking shouldbe informed by the Netherlands Institute of Mental Health andAddiction about the importance of implementing the com-bined PAS intervention aimed at both students and theirparents. Adolescents in lower levels of education and innon-Islamic school should be a specific target when imple-menting the PAS intervention. However, since no adverseeffects were found in other groups of adolescents, PAS can beimplemented in schools with adolescents from different reli-gious and educational backgrounds. The present findings high-light the conclusion that the separate interventions should notbe implemented. Even in cases where adolescents have highself-control and/ or parents have strict rules, the separate inter-ventions are not shown to be effective, thereby indicating thatan increase in these behaviors only is not sufficient to fosterchange. Prevention workers should be kept updated about newinsights and knowledge that are obtained in the future.

Acknowledgements This study was funded by grant number 62200021 from the Dutch Health Care Research Organization (Z.O.N.-M.W). We would like to thank the Associate Editor and the three anon-ymous reviewers who provided us with thoughtful comments on therevision.

Conflict of interest The authors declare that there are no conflicts ofinterest.

Open Access This article is distributed under the terms of the Crea-tive Commons Attribution Noncommercial License which permits anynoncommercial use, distribution, and reproduction in any medium,provided the original author(s) and source are credited.

References

Ajzen, I., & Fishbein,M. (1990).Understanding attitudes and predictingsocial behaviour. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1996). Social foundation of thought and actions: A socialcognitive theory. Englewood Cliffs: Prentice Hall.

Behrendt, S., Wittchen, H. U., Höfler, M., Lieb, R., & Beesdo, K.(2009). Transitions from first substance use to substance usedisorders in adolescence: Is early onset associated with a rapidescalation? Drug and Alcohol Dependence, 99, 68–78.

Brody, G. H., Flor, D. L., Hollett-Wright, N., McCoy, J. K., & Donovan,J. (1999). Parent–child relationships, child temperament profiles andchildren’s alcohol use norms. Journal of Studies on Alcohol, 13,45–51.

Brody, G. H., Kogan, S. M., Yi-fu, C., & McBride Murry, V. (2008).Long-term effects of the Strong African American Families Programon youths’ conduct problems. Journal of Adolescent Health, 43,474–481.

Brown, S. A., Anderson, K. G., Schulte, M. T., Sintov, N. D., &Frissell, K. C. (2005). Facilitating youth self-change throughschool-based intervention. Addictive Behaviors, 30, 1797–1810.

286 Prev Sci (2012) 13:278–287

Brown, S. A., Carello, P., Vik, P. W., & Porter, R. J. (1998). Changes inalcohol and self-efficacy and alcohol expectancies during addic-tion intervention. Substance Abuse, 19, 155–167.

Brown, S. A., McGue, M., Maggs, J., Schulenberg, J., Hingson, R.,Swartzwelder, S., et al. (2008). A developmental perspective onalcohol and youths 16 to 20 years of age. Pediatrics, 121, S290–S310.

Del Boca, F. K., & Darkes, J. (2003). The validity of self-reports ofalcohol consumption: State of the science and challenges forresearch. Addiction, 98, 1–12.

Demmel, R., Beck, B., Richter, D., & Reker, T. (2004). Readiness tochange in a clinical sample of problem drinkers: Relation to alcoholuse, self-efficacy, and treatment outcome. European AddictionResearch, 10, 133–138.

Engels, R. C. M. E., & Knibbe, R. A. (2000). Alcohol use and intimaterelationships in adolescence: When love comes to town. AddictiveBehaviors, 25, 435–439.

Engels, R. C. M. E., Knibbe, R. A., & Drop, M. J. (1999). Why do lateadolescents drink at home? A study on the psychological well-being, social integration and drinking context. Addiction Research,7, 31–46.

Foxcroft, D.R., Ireland, D., Lowe, G., & Breen, R. (2002). Pri-mary prevention for alcohol misuse in young people.Cochrane Database of Systematic Reviews. 3, Art. No.:CD003024.

Hibell, B., Guttormsson, U., Ahlström, S., Balakireva, O., Bjarnason,T., Kokkevi, A., & Kraus, L. (2009). The 2007 ESPAD report -substance use among students in 35 European countries. Stock-holm: The Swedish Council for Information on Alcohol and OtherDrugs (CAN).

Koning, I. M., Engels, R. C. M. E., Verdurmen, J. E. E., & Vollebergh,W. A. M. (2010). Alcohol-specific socialization practices andalcohol use in Dutch early adolescents. Journal of Adolescence,33, 93–100.

Koning, I. M., Harakeh, Z., Engels, R. C. M. E., & Vollebergh, W. A.M. (2010). A comparison of self-reported alcohol use measuresby early adolescents: Questionnaires versus diary. Journal ofSubstance Use, 15, 166–173.

Koning, I. M., van den Eijnden, R. J. J. M., Engels, R. C. M. E.,Verdurmen, J. E. E., & Vollebergh, W. A. M. (2010). Why targetearly adolescents and parents in alcohol prevention? The mediat-ing effects of self-control, rules and attitudes about alcohol use.Addiction, 106, 538–564.

Koning, I. M., Vollebergh, W. A. M., Smit, F., Verdurmen, J. E.E., van den Eijnden, R. J. J. M., ter Bogt, T. F. M., et al.(2009). Preventing heavy alcohol use in adolescents (PAS):Cluster randomized trial of a parent and student interventionoffered separately and simultaneously. Addiction, 104, 1669–1678.

Koutakis, N., Stattin, H., & Kerr, M. (2008). Reducing youth alcoholdrinking through a parent-targeted intervention: The Örebro Pre-vention Program. Addiction, 103, 1629–1637.

Kraemer, H. C., Wilson, G. T., Fairburn, C. G., & Agras, W. S.(2002). Mediators and moderators of treatment effects inrandomized clinical trials. Archives of General Psychiatry, 59,877–883.

Monshouwer, K., Smit, F., De Zwart, W. M., Spruit, I., & VanAmeijden, E. J. C. (2003). Progress from a first drink to firstintoxication: Age of onset, time-windows and risk factors in aDutch national sample of secondary school students. Journal ofSubstance Use, 8, 155–163.

Monshouwer, K., Verdurmen, J., van Dorsselaer, S., Smit, E., Gorter,A., & Vollebergh, W. (2008). Jeugd en riskant gedrag 2007

[Adolescents and risk-taking behaviour 2007]. Utrecht, The Neth-erlands: Trimbos-instituut.

Salonna, F., van Dijk, J.P., Geckova, A. M., Sleskova, M., Groothoff, J.W., & Reijneveld, S. A. (2008). Social inequalities in changes inhealth-related behavior among Slovak adolescents aged between15 and 19: A longitudinal study. BMC Public Health, 8.

Sloboda, Z., Stephens, R. C., Stephens, P. C., Grey, S. F., Teasdale, B.,Hawthorne, R. D., et al. (2009). The adolescent substance abuseprevention study: A randomized field trial of a universal sub-stance abuse prevention program. Drug and Alcohol Dependence,102, 1–10.

Smit, E., Verdurmen, J., Monshouwer, K., & Smit, F. (2008). Familyinterventions and their effect on adolescent alcohol use in generalpopulations: A meta-analysis of randomized controlled trials.Drug and Alcohol Dependence, 97, 195–206.

Spijkerman, R., van den Eijnden, R. J., & Huiberts, A. (2008). Socio-economic differences in alcohol-specific parenting practices andadolescents’ drinking patterns. European Addiction Research, 14,26–37.

Spoth, R., Greenberg, M., & Turrisi, R. (2008). Preventive interven-tions addressing underage drinking: State of the evidence andsteps toward public health impact. Pediatrics, 121, 311–336.

Spoth, R. L., Randall, G. K., Trudeau, L., Shin, C., & Redmond, C.(2008). Substance use outcomes 51/2 years past baseline forpartnership-based family-school preventive interventions. Drugand Alcohol Dependence, 96, 57–68.

Spoth, R. L., Shin, C., Guyll, M., Redmond, C., & Azevedo, K. (2006).Universality of effects: An examination of the comparability oflong-term family intervention effects on substance use across risk-related subgroups. Prevention Science, 7, 209–224.

Sumter, S. R., Bokhorst, C. L., Steinberg, L., & Westenberg, P. M.(2009). The developmental pattern of resistance to peer influencein adolescence: Will the teenager ever be able to resist? Journal ofAdolescence, 32, 1009–1021.

Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High self-control predicts good adjustment, less pathology, better grades,and interpersonal success. Journal of Personality, 72, 271–322.

Tobler, N. S., Roona, M. R., Ochshorn, P., Marshall, D. G., Streke, A.V., & Stackpole, K. M. (2000). School-based adolescent drugprevention programs: 1998 meta-analysis. Journal of PrimaryPrevention, 20, 275–336.

Trudeau, L., Spoth, R., Lillehoj, C., Redmond, C., & Wickrama, K. A.S. (2003). Effects of a preventive intervention on adolescentsubstance use initiation, expectancies, and refusal skills. Preven-tion Science, 4, 109–122.

Van der Vorst, H., Engels, R. C. M. E., Dekovic, M., Meeus, W., &Vermulst, A. A. (2007). Alcohol-specific rules, personality andadolescents’ alcohol use: A longitudinal person-environmentstudy. Addiction, 102, 1064–1075.

Van der Vorst, H., Engels, R. C. M. E., Meeus, W., & Dekovic,M. (2006). The impact of alcohol-specific rules, parentalnorms about early drinking and parental alcohol use onadolescents’ drinking behavior. Journal of Child Psychologyand Psychiatry, 47, 1299–1306.

Van der Vorst, H., Engels, R. C. M. E., Meeus, W., Dekovic, M., & VanLeeuwe, J. (2005). The role of alcohol-specific socialization inadolescents’ drinking behaviour. Addiction, 100, 1464–1476.

Wills, T. A., Cleary, S., Filer, M., Shinar, O., Mariani, J., & Spera, K.(2001). Temperament related to early-onset substance use: Test ofa developmental model. Prevention Science, 2, 145–163.

Yu, J. (2003). The association between parental alcohol-relatedbehaviors and children’s drinking. Drug and Alcohol Dependence,69, 253–262.

Prev Sci (2012) 13:278–287 287